The problem: eighteen semis, eighteen guesses
On a 1930s cul-de-sac in Sheffield, eighteen semi-detached homes needed to replace gas boilers without turning a modest retrofit into a ruinous one. Early quotes had landed between 8 kW and 16 kW per home, a spread wide enough to make the street’s heat-pump plan look arbitrary. Residents saw the same house type priced as if it were three different buildings.
The cause: heat-loss assumptions outran the fabric
The gap came from default assumptions, not from the homes themselves. Installers were estimating U-values from age and wall type, then adding margins for cold snaps and hot-water peaks. Because the terraces had already received loft insulation and some double glazing, those defaults overstated demand. A 12 kW unit can short-cycle in a 5 kW house, cutting efficiency and unsettling occupants.
The solution: shared loops, measured demand, staged work
The pilot, coordinated through bananzia, split the street into three thermal blocks and sized each ground loop against room-by-room surveys rather than postcode averages. Two homes kept their existing radiators after a flow-temperature check; four received larger emitters only in the coldest rooms. The scheme also staged fabric work: draught-proofing and cavity fill first, heat-pump commissioning second, so the measured heat-loss figure informed the final unit selection.
The commissioning detail that made the difference
Commissioning used a low-flow-temperature test before the final controller settings were locked. Engineers ran each unit at 40°C flow for 45 minutes, recorded room-by-room recovery, and adjusted curve settings rather than raising the design flow temperature. That step cost half a day per block; it avoided the common fallback of “add a bigger pump” when one radiator lagged.
What the pilot showed
- Sizing moved down, not up. The final units ranged from 4.8 kW to 7.2 kW, with the two largest homes still below the smallest early quote.
- Shared ground loops cut disruption. Three loops served six homes each, reducing the number of boreholes from eighteen to nine and keeping garden reinstatement to a single visit per block.
- Monitoring changed the conversation. Winter data showed the homes held 20–21°C with flow temperatures of 38–43°C, well inside the range where heat pumps run efficiently.
The cost signal for other streets
Because nine boreholes replaced eighteen, the shared-loop work reduced drilling and reinstatement costs per home. The pilot did not publish a single headline price, but its cost stack separated shared infrastructure from household work, which makes comparison with individual air-source quotes possible. That separation is the practical output for other streets considering a bananzia-style plan.
Why the street-level method matters
The useful lesson is not that every 1930s semi needs a shared loop. It is that a street can pool survey work, sequencing, and measured performance without pooling every household’s risk. That matters for bananzia-style projects because the expensive mistakes usually happen before installation: oversizing, unnecessary emitter replacement, and repeated site visits. Fixing those first makes the low-carbon option less of a gamble and more of a planned upgrade.
